metrics for effort/cost estimation of mobile apps development
TRANSCRIPT
Università degli studi di SalernoDipartimento di Scienze Aziendali, Management & Innovation SystemCorso di Laurea Magistrale in Tecnologie Informatiche e Management
Metrics for Effort/Cost Estimation of Mobile apps development
ANNO ACCADEMICO 2015-2016
Relatore: Prof. ssa Filomena FerrucciDott. Pasquale Salza
Candidata: Catolino Gemma
Matricola 0222500095
Tesi di laurea magistrale in Ingegneria del Software: Metriche, Qualità e Valutazione Sperimentale
The Effort and Cost Estimation
Not Only for traditional Software
always hard to estimate in advance
over budget
and overrun
Continuous process
When more data become available…
more accurate estimations can be achieved!
Non-Model-Based
Human experts
MAN / HOURS
Model-BasedM M M M
MAN / HOURS
the Size Factor
the Size Factor
L OC
L OC
the Size Factor
F P AFunction Point Analysis
(functional) transactions
and (logical) data
the Size Factor
c f pCosmic Function Point
movements from/to persistent
storage and users
the Size Factor
D’avanzo et al. approach
van Heeringen & van Gorp
approach
Sellami et al.
Set of guidelines for
an approximate and
quick sizing of mobile apps
IFPUG Guidelines
D’avanzo approach
van Heeringen & van Gorp
approach
Cozzolino et al. approach
new set of guidelines
Cozzolino et al. approach
View/Show Data
Create/Set/Delete Data
Invoking service
new set of guidelines
Cozzolino et al. approach
3 CFP
3 CFP
2 CFP
new set of guidelines
LIMITations
The software life cycle is already started!
Early Effort Estimation
Defining a set of metrics for mobile early effort estimation
Defining a set of metrics for mobile early effort estimation
Investigating how the early size measure can be mapped
into Cozzolino et al. guidelines
Defining a set of metrics for mobile early effort estimation
Investigating if the mapping is useful for estimating CFP
Investigating how the early size measure can be mapped
into Cozzolino et al. guidelines
Defining a set of metrics for mobile early effort estimation
Emilia Mendes
Emilia Mendes, Nile Mosley, and Steve Counsell.Investigating early web size measures for web cost estimation. In Proceedings of EASE’2003 Conference,Keele, pages 1–22, 2003.
Analysis of quote form
Emilia Mendes, Nile Mosley, and Steve Counsell.Investigating early web size measures for web cost estimation. In Proceedings of EASE’2003 Conference,Keele, pages 1–22, 2003.
Emilia Mendes, Nile Mosley, and Steve Counsell.Investigating early web size measures for web cost estimation. In Proceedings of EASE’2003 Conference,Keele, pages 1–22, 2003.
377manually validated links
Analysis of quote form
Analysis of quote form
Emilia Mendes, Nile Mosley, and Steve Counsell.Investigating early web size measures for web cost estimation. In Proceedings of EASE’2003 Conference,Keele, pages 1–22, 2003.
Extraction of initial set of metrics
Features
Categories
Features Application GUI
Categories
Features Application GUICost
Driver
Categories
Features Application GUI
Project’s Metrics
Cost Driver
Categories
Features Application GUI
Project’s Metrics
Cost Driver
Application functionality
Categories
Features Application GUI
Project’s Metrics
Cost Driver
Application functionality
Application size
Categories
Features Application GUI
Project’s Metrics
Cost Driver
Application functionality
Possible Metrics
Application size
Categories
Emilia Mendes, Nile Mosley, and Steve Counsell.Investigating early web size measures for web cost estimation. In Proceedings of EASE’2003 Conference,Keele, pages 1–22, 2003.
Extraction of initial set of metrics Validation of initial
set of metrics
Analysis of quote form
Validation of initial set of metrics
42
DEVELOPERS
PROJECT MANAGERS
Validation of initial set of metrics
TWO SURVEYS
Validation of initial set of metrics
TWO SURVEYS
Validation of initial set of metrics
TWO SURVEY
YY
Validation of initial set of metrics
48 METRICS
Validation of initial set of metrics
36CONFIRMED
Validation of initial set of metrics
12DELETED
Validation of initial set of metrics
12DELETED
Project start date
App purchasing
Type of business owns the app idea
Complex back-end
Validation of initial set of metrics
5 ADDED
Validation of initial set of metrics
5 ADDED
Support Security
Backward compatibility
User target
Features
Generalities
Projects Design
Platfom
Accounting
User featuresSocial Aspect
Remote Connection
eCommerce
Date & Location
MonitoringAdditional
Functionality
Renovation of Categories
DesignProjects
Platfom
Features
Date & Location
MonitoringAdditional
Functionality
Renovation of Categories
Accounting
User featuresSocial Aspect
Generalities
Remote Connection
eCommerceGoogle Module
SIZE
Investigating how the early size measure can be mapped into Cozzolino et al. guidelines
Requirements
Early phase of developments Requirement Elicitation/ Analysis
SOFTWARE SIZE
Early phase of developments Requirement Elicitation/ Analysis
Early phase of developments Requirement Elicitation/ Analysis
cosmic
View/Show Data
Exchange Data via a network
Invoking service
Create/Set/Delete Data
GuidelinesCozzolino et al.
Early MetricsSocial sharing
Search
MessagingAd hoc
authentication
Analytics
Exchange Data via a network
Early Metrics
Ad hoc authentication
GuidelinesCozzolino et al.
Exchange Data via a network
Lines guideCozzolino et al.Early Metrics
Ad hoc authentication
MIN MAX10 CFP5 CFP
Login + RegisterLogin
Ad hoc authentication
Exchange Data via a network
Lines guideCozzolino et al.Early Metrics
Ad hoc authentication
MINMAX10 CFP 5 CFP
Login + Register Login
41 METRICS
Exchange Data via a network
Lines guideCozzolino et al.Early Metrics
Ad hoc authentication
MINMAX10 CFP 5 CFP
Login + Register Login
26 METRICSMIN MAXOPERATIONS
Empirical study
Evaluate the accuracy of the estimations in terms of
COSMIC Function Pointsof the early metrics
RQ: To what extent the CFPs extractable using the early metrics are close to the actual CFPs of a Mobile app?
Evaluate the accuracy of the estimations in terms of
COSMIC Function Pointsof the early metrics
Evaluate the accuracy of the estimations in terms of
COSMIC Function Pointsof the early metrics
13 MOBILE APPLICATIONS
APP FUR EARLYMETRIC
DESIGN
EARLYMETRIC
#CFP
DESIGN
MIN
MAXAVG
EARLYMETRIC
#CFP
MRE
MMRE
MdMRE
PRED(25)
DESIGN
Results
Application Early CFP_min Early CFP_max Early CFP_avg Oracle
Wikipedia 37 47 42 46
Munch 41 51 46 42Loopboard 16 21 18,5 14
Man man 34 44 38,5 38
Easy Sound Recorder 20 25 22,5 18
K-9 Mail 38 53 45,5 32
Transportr 47 67 57 38
Hashr 23 23 23 19
arXiv Mobile 37 42 39,5 39NPR News 37 42 39,5 38
Loop Habit Tracker 26 31 28,5 28
Radio Droid 33 38 35,5 50
RoomMates Expense 26 31 28,5 44
Application Early CFP_min Early CFP_max Early CFP_avg Oracle
Wikipedia 37 47 42 46
Munch 41 51 46 42Loopboard 16 21 18,5 14
Man man 34 44 38,5 38
Easy Sound Recorder 20 25 22,5 18
K-9 Mail 38 53 45,5 32
Transportr 47 67 57 38
Hashr 23 23 23 19
arXiv Mobile 37 42 39,5 39NPR News 37 42 39,5 38
Loop Habit Tracker 26 31 28,5 28
Radio Droid 33 38 35,5 50
RoomMates Expense 26 31 28,5 44
Application Early CFP_min MRE_min PRED(25)
Wikipedia 37 0,19 1
Munch 41 0,02 1Loopboard 16 0,14 1
Man man 34 0,01 1
Easy Sound Recorder 20 0,11 1
K-9 Mail 38 0,19 1
Transportr 47 0,24 1
Hashr 23 0,21 1
arXiv Mobile 37 0,05 1NPR News 37 0,03 1
Loop Habit Tracker 26 0,07 1
Radio Droid 33 0,34 0
RoomMates Expense 26 0,41 0
MIN
MIN
MMRE 0,16
MDMRE 0,14
PRED(25) 85%
Application Early max MRE_max PRED(25)
Wikipedia 47 0,02 1
Munch 51 0,21 1Loopboard 21 0,5 0
Man man 44 0,16 1
Easy Sound Recorder 25 0,39 0
K-9 Mail 53 0,66 0
Transportr 67 0,76 0
Hashr 23 0,21 1
arXiv Mobile 42 0,08 1NPR News 42 0,1 1
Loop Habit Tracker 31 0,11 1
Radio Droid 38 0,24 1
RoomMates Expense 31 0,29 1
MAX
MIN
MMRE 0,29 0
MDMRE 0,21
PRED(25) 61%
Application Early avg MRE_avg PRED(25)
Wikipedia 42 0,09 1
Munch 46 0,09 1Loopboard 18,5 0,32 0
Man man 38,5 0,01 1
Easy Sound Recorder 22,5 0,25 1
K-9 Mail 45,5 0,42 0
Transportr 57 0,5 0
Hashr 23 0,21 1
arXiv Mobile 39,5 0,01 1NPR News 39,5 0,04 1
Loop Habit Tracker 28,5 0,02 1
Radio Droid 35,5 0,29 0
RoomMates Expense 28,5 0,35 0
AVG
MMRE 0.2
MDMRE 0.21
PRED(25) 61%
AVG
Application Early CFP_min Early CFP_max Early CFP_avg Oracle
Wikipedia 37 47 42 46
Munch 41 51 46 42Loopboard 16 21 18,5 14
Man man 34 44 38,5 38
Easy Sound Recorder 20 25 22,5 18
K-9 Mail 38 53 45,5 32
Transportr 47 67 57 38
Hashr 23 23 23 19
arXiv Mobile 37 42 39,5 39NPR News 37 42 39,5 38
Loop Habit Tracker 26 31 28,5 28
Radio Droid 33 38 35,5 50
RoomMates Expense 26 31 28,5 44
Application Early CFP_min Early CFP_max Early CFP_avg Oracle
Wikipedia 37 47 42 46
Munch 41 51 46 42Loopboard 16 21 18,5 14
Man man 34 44 38,5 38
Easy Sound Recorder 20 25 22,5 18
K-9 Mail 38 53 45,5 32
Transportr 47 67 57 38
Hashr 23 23 23 19
arXiv Mobile 37 42 39,5 39NPR News 37 42 39,5 38
Loop Habit Tracker 26 31 28,5 28
Radio Droid 33 38 35,5 50
RoomMates Expense 26 31 28,5 44
RQ: To what extent the CFPs extractable using the early metrics are close to the actual CFPs of a Mobile app?
The estimations provided by our metrics resulted quite
close to the actual values
EARLYMETRIC
Additional validation with companies
Gather data
FUTURE WORK
FUTURE WORK
EARLYMETRIC CFP+
Summary
Summary
Summary
Summary
Summary
Summary
Thank you!